Customer classification: A Mamdani fuzzy inference system standpoint for modifying the failure mode and effect analysis based three dimensional approach

Abstract

Customer categorization using a three dimensional loyalty matrix, based on failure mode and effect analysis (FMEA), is an innovative approach for customer classification but is vulnerable to FMEA limitations. The main purpose of this research is to utilize a multi input single output Mamdani fuzzy inference system (FIS) to cope with the traditional FMEA inherited shortcomings. Besides, the classification logic and classes of the Loyalty Matrix methodology have been adopted for the purpose. We have also identified four potential market scenarios and evaluated the performance of the proposed methodology within these contexts. Correspondingly, four tailored FIS’s consisting of a total of 108 fuzzy rules have been developed. Empirical results indicate that the new approach successfully resolved serious issues such as data uncertainty, weight ignorance, the same output value computation from different input values and the discontinued output.